Maximum Current Specs

[johnl]

The data sheet for a PIC16F720 says:

Maximum current out of VSS pin ......................... 95 mA
Maximum current sunk by all ports, -40°C to +85°C for
industrial............ 200 mA

How can ports sink 200mA without exceeding 95mA out of Vss?

It's a similar scenario for sourcing and current into Vdd.

What am I missing?

[asmboy]

i have always treated the LESSER value as steady state MAX DC value.

however for very short pulses - you can safely come closer to the HIGHER of the values. it is a fool's game to depend on any value greater than the MAX current into or out of any ONE pin-(+/-25ma) or to exceed the maximum core Vss Vdd specs on the sheet.

if you need lots of current -use an external transistor or mosfet to do the heavy lifting. that's what the pros do.

[johnl]

I guess I'm questioning the logic of the specs. If they say you can't have more than 95 mA out of the Vss pin, then they should say you can't have more than 95 mA sunk by all ports.

[PCM programmer]

I think that data sheet is in error. The 16F887 data sheet says:

[johnl]

Yes, it does look like an error, though there doesn't seem to be an errata sheet that mentions it.

I have a design that will sink 85mA into the the C port, but I'd feel better if I knew the 95 mA spec out of the Vss pin is supposed to be higher.

I'll write to Microchip.

[dyeatman]

My advice is to avoid sinking that kind of current with the PIC regardless.
You are begging for current transient issues. Why not use a FET driver
chip (ULNxxxx) instead? Cheap, much more stable and adds protection
for the PIC as a bonus. Not to mention heating issues.
_________________
Google and Forum Search are some of your best tools!!!!

[RF_Developer]

[temtronic]

I have to 'second' dyeatman about using a ULNdriver chip ! Something rarely discussed is overheating of PICs and well, bad things happen when they get hot. If one region of the PIC is hotter ten the clock can go wrong, ADC goes whacky, I/O latches up, etc. There is a 'max temp, heat disp' spec you need to be WELL under in order to be sure the PIC is 'happy' and will outlive you and I.

It's well worth the extra buck per PCB to add the ULN,even add a socket for the 'high powered' driver. My energy control boards (4000 series, pre PIC) had every chip in sockets.Though it cost me a penny a pin,repair time to swap out a bad chip was FAST and easy unlike desoldering,cleaning, install,cleaning,etc.

Also be sure the PCB traces for the 'high power' section are wide enough to handle the current! I see too many designs with ultra thin traces that need to pump 1/2 amp through ! Those usually become unknown 'onboard fuses' !!

Jay

[johnl]

This application needs to be cheap. Sinking 12 mA into each of 7 I/O pins doesn't seem to be asking too much, given the fact that each can handle 25 mA and the feature sheet boasts direct LED drive.

Aside from powering an LED display, the app is basically static, just checking a few switches and sourcing a few other LEDs. No time-sensitive peripherals are used and no A/D.

Is it wrong to assume it will be fine since total dissipation will be less than 60 or 70 mW? (maximum is 800 mW.)

The current out of the Vss pin is the only rating I'd be pushing and is my only concern.

I have a ticket into Microchip and will post their response.

Thanks for the responses.

[asmboy]

[temtronic]

re:
This application needs to be cheap.

Be careful !! saving a few pennies now may spell disaster in the field !! Just cause it 'worked fine on the bench' doesn't mean it will for the customer. BTDT. Had an RC osc setup worked fine until the temperature went up and humdidty set in. Timing went 'out the window'....sigh, had to drop xtal and 3 chips in to get the right frequency.BTW nobody made 24Hz xtals 20 years ago......

also.. you might consider low current LEDs,if you haven't yet. Some are great at 5ma !

Jay

[Ttelmah]

The 720, is very specifically a low power chip. The current spec given is for the full temperature range. Maximum source current is only 65mA over this range. Beware also, that the Voh specifications only go up to 3.5mA for the pins. They droop a _lot_ more as the current goes above this, result more dissipation....
The pull down drivers have lower droop, so reduce dissipation. Better to work this way if you can.
So the question becomes what is the maximum temperature range you need (remember it can be surprising the temperature that quite mundane circuitry will meet...).
Chips like the ULN2003, in SMD, come in at just $0.13 in sensible quantities.
Remember with high efficiency LED's, 2.5mA is a fairly typical current, so 'direct LED drive' does not imply high current capability.

[johnl]

The spec for the current out of the Vss pin might be an error, as PCM programmer mentioned. (I am waiting to hear from Microchip on that.) So for the purposes of this discussion, let's just look at the other limits.

The consensus seems to be very conservative here and I don't see why that is.

The chip is capable of sinking 25 mA from each I/O pin and I'm looking to sink 12 mA from 7 of them in a modest temperature range environment of 0 deg.C to 50 deg. C.

So the current sunk by all ports is 84mA -less than half of the 200 mA rating.

If Vol is maybe .8v at 12 mA, then the power is about 70mW.

Why are running a dissipation of one tenth the max power and output pin current of one half the max and total sink current less than one half the max causes for concern?

To what margin do you guys normally design? I thought 50% of max has been a useful rule of thumb.

[dyeatman]

The reason for my caution is extensive experience. When you start
switching loads directly with the PIC you get into switching transients/EMF.
I have designed many systems (PIC and other) and I always design on the
conservative side to maintain reliability.
_________________
Google and Forum Search are some of your best tools!!!!

[temtronic]

Have you tested in 'real life' yet? 0*C is easy, toss into a freezer.. 50*C gets to be more fun, though I have built small 'environmental test chambers' in the past. You might need the 'industrial' rated PICs not the cheaper 'commercial' ones but check the specs real carefully and do some honest testing BEFORE you ship out 10,000 units and have half come back because they 'don't work right'.

Jay